Presented by CACI
As artificial intelligence becomes more deeply integrated into intelligence operations, agencies face a challenge that extends well beyond acquiring new technology: preparing their people to apply AI effectively to real-world missions.
Dave Reynolds, Senior Vice President and Division Manager for Mission and Engineering Support at CACI, says developing AI literacy and fluency across the intelligence workforce is one of the company’s primary areas of focus. But that effort should not revolve around teaching analysts how to operate a particular product. Instead, it should help them understand how AI techniques can be applied within intelligence tradecraft to achieve specific mission outcomes.
That distinction is especially important because tools, models and available data are constantly changing. Training tied too closely to a single platform can quickly become outdated. A tradecraft-centered approach gives analysts skills they can transfer across tools, agencies and mission environments.
“The ability to free up their thinking allows them to be much better analysts,” Reynolds said.
AI fluency, however, will not look identical across every intelligence discipline. A geospatial analyst conducting pattern-of-life analysis works with different data flows and mission requirements than an analyst examining political behavior, motivations or likely future actions.
In geospatial intelligence, AI may be particularly effective at correlating large data sets across time and location. Political or predictive analysis may require greater attention to context, intent and other factors that depend heavily on human judgment. The underlying concept of AI fluency applies to both, but its practical application must be tailored to each domain.
Reynolds describes AI as a skill that analysts apply rather than simply a tool they use. That means beginning with the mission outcome, understanding the customer’s challenges and then determining how AI can improve the workflow. The approach will differ across organizations such as the Defense Intelligence Agency and National Geospatial-Intelligence Agency because each has distinct responsibilities, data and operational requirements.
This mission-centered approach also reinforces the importance of intelligence tradecraft. Analysts must understand how information is generated, validated and transformed into products that senior leaders can use to make decisions. AI-generated speed has limited value if the resulting information is inaccurate, irrelevant or unsuitable for operational use.
For Reynolds, successful implementation starts with understanding the desired decision advantage and working backward to build the appropriate analytic process.
Open architectures are another essential part of that process. Intelligence agencies increasingly want to avoid becoming locked into individual platforms or proprietary frameworks. They need solutions that can work across structured and unstructured data, accommodate changing data sources and share information across multiple domains and mission partners.
Reynolds pointed to dark-web exploitation as an example. Information in the deep and dark web changes rapidly, and the value of that information can vary by mission and customer. A tool-agnostic, open approach allows collected data and analytic results to fit within the government’s existing frameworks and be shared appropriately across intelligence partners.
Cybersecurity and data authentication must remain central to this work. The intelligence community relies increasingly on publicly available information, but analysts must verify that the information is accurate and has not been manipulated. Agencies must also protect their data and analytic processes from adversaries seeking to access or exploit them.
The proliferation of open-source information has profoundly changed intelligence analysis, according to Reynolds. Analysts can now draw on a much broader range of information to understand events, identify patterns and evaluate potential threats. The challenge is connecting that material with classified holdings to create a holistic intelligence picture.
This evolution is also changing how agencies train their workforces. Analysts need the skills to assess open-source information, authenticate it, combine it with other holdings and apply it responsibly to mission questions.
Policy has not always kept pace with the technology or the expanding availability of data. Reynolds believes closer government-industry partnerships can help agencies move more quickly from studying these challenges to executing practical solutions.
The ultimate objective is not to remove the analyst from the intelligence process. It is to use AI to reduce the time analysts spend collecting and organizing information so they can devote more attention to interpretation, critical thinking and decisions. By grounding AI fluency in mission outcomes, open architectures and established intelligence tradecraft, agencies can strengthen their workforces while preserving the human judgment at the center of the mission.